Skeleton-aided Articulated Motion Generation

TL;DR

This paper introduces a novel method for generating realistic articulated human motion sequences from a single image by combining skeleton information and appearance references using a conditional GAN and triplet loss.

Contribution

It is the first to generate articulated motion sequences from a single image leveraging skeleton and appearance data within a GAN framework.

Findings

Produces realistic articulated motion sequences

Outperforms previous methods in motion realism

Effective on KTH and Human3.6M datasets

Abstract

This work make the first attempt to generate articulated human motion sequence from a single image. On the one hand, we utilize paired inputs including human skeleton information as motion embedding and a single human image as appearance reference, to generate novel motion frames, based on the conditional GAN infrastructure. On the other hand, a triplet loss is employed to pursue appearance-smoothness between consecutive frames. As the proposed framework is capable of jointly exploiting the image appearance space and articulated/kinematic motion space, it generates realistic articulated motion sequence, in contrast to most previous video generation methods which yield blurred motion effects. We test our model on two human action datasets including KTH and Human3.6M, and the proposed framework generates very promising results on both datasets.